Able to create volume, but unable to mount it to pod about csi-digitalocean

[JoschaP]

+1

[fatih]

This seems to be an issue with your k8s version. It's very old and not supported by this driver (the error failed to attach: node "openshift-master" has no NodeID annotation is because the node couldn't be registered to be used by CSI) .

Please upgrade it to v1.12.1 at least minimum. Happy to debug if it's still happening for you.

[JoschaP]

The fix was to DOWNgrade from 0.3.x to 0.2.0 and it worked instantly

[fatih]

@JoschaP care to share what you have found? The 0.3.1 has a some serious fixes so it should work nevertheless. You should't have to downgrade

[badri]

@fatih thanks for your response man. OKD always runs with the corresponding version of Kubernetes. i.e. I use OKD 3.10 and it uses 1.10 version of Kubernetes.

Source: https://docs.okd.io/3.10/architecture/infrastructure_components/kubernetes_infrastructure.html#overview

Current latest stable is OKD 3.11 and it might take a while before it bumps up to 3.12. Is there any way I can backport the updates to Kubernetes 3.10? I can own that task if you tell me what needs to be done and contribute it back to the code base.

I want OKD to work with DigitalOcean.

[badri]

Also, I noticed that I use 0.2.0 for my setup which is compatible with Kubernetes 1.10.

image: digitalocean/do-csi-plugin:v0.2.0
Any idea why it still might not work?

[JoschaP]

@fatih I can't exactly tell you why it worked then. We tried it before with the 0.3.1 but couldn't get it to work. Out of "desperation" we simply installed 0.2.0 and it worked immediately.

The error with the 0.3.1 was that it executed and mounted the VolumeClaim correctly, but the Pod did not come up. With the 0.2.0 it worked as said immediately.

We are using Kubernetes v1.11.5-rancher1-1

[fatih]

@badri unfortunately v1.10 is not suggested because it has some serious bugs in day to day operating of volumes. You might create a workload with a volume attached to it, however, you might see problems later when you upgrade the pod. Hence we don't suggest using v1.10 at all. Also the minimum version is v1.10.5 for CSI v0.2.0. The v1.10.0 is not compatible with it.

[fatih]

@JoschaP I'm happy to help and debug your 0.3.1 cluster if you can create one. Just let me know how we can proceed.

[dbogatov]

I can confirm the same problem on v0.3.1 and the downgrade to v.0.2.0 fixed everything.

[fatih]

v0.3.1 is not compatible with v1.11.x, it's only compatible with v.1.12.x. If you se an old version you need to use v0.2.0.

[badri]

I now use 1.11, but still get the same issue.

$ oc version
oc v3.11.0+62803d0-1
kubernetes v1.11.0+d4cacc0
features: Basic-Auth GSSAPI Kerberos SPNEGO

Server https://openshift-master:8443
openshift v3.11.0+92b7c41-132
kubernetes v1.11.0+d4cacc0